Objective Since WHO released the package of six MPOWER measures to assist nations with implementing the WHO Framework Convention for Tobacco Control (FCTC), 88 countries adopted at least one highest level MPOWER measure. We estimated the subsequent reduction in smoking-related deaths from all new highest level measures adopted between 2007 and 2014.
Methods Policy effect sizes based on previously validated SimSmoke models were applied to the number of smokers in each nation to determine the reduction in the number of smokers from policy adoption. On the basis of research that half of all smokers die from smoking, we derived the smoking-attributable deaths (SADs) averted of those smokers alive today.
Findings In total, 88 countries adopted at least one highest level MPOWER policy between 2007 and 2014, resulting in almost 22 million fewer projected SADs. The largest number of future SADs averted was due to increased cigarette taxes (7.0 million), followed by comprehensive smoke-free laws (5.4 million), large graphic health warnings (4.1 million), comprehensive marketing bans (3.8 million) and comprehensive cessation interventions (1.5 million).
Conclusions These findings demonstrate the immense public health impact of tobacco control policies adopted globally since the WHO-FCTC and highlight the importance of more countries adopting highest level MPOWER measures to reduce the global burden of tobacco use. Substantial additional progress could be made, especially if heavily populated nations with high smoking prevalence were to reach highest level MPOWER measures.
- Public policy
- Smoking Caused Disease
- Global health
- Low/Middle income country
Statistics from Altmetric.com
Smoking-attributable deaths (SADs) are projected to rise to eight million deaths annually worldwide by 2030,1 due primarily to growth in smoking in low and medium income countries (LMICs). The WHO Framework Convention on Tobacco Control (FCTC) is the first international global health treaty and seeks to reduce this burden by providing countries with a legally binding instrument to guide impactful tobacco control policy measures. To assist nations with implementing WHO-FCTC obligations, WHO released the MPOWER package of six evidence-based WHO-FCTC tobacco control measures in 2008. The MPOWER package includes: Monitoring tobacco use and tobacco control measures; Protecting people from tobacco smoke; Offering help (eg, treatments) to quit tobacco; Warning people about the dangers of tobacco; Enforcing bans on tobacco advertising, promotion and sponsorship and Raising tobacco taxes.
While the WHO-FCTC came into force in 2005, global tobacco control efforts have been accelerated since WHO first introduced the MPOWER package in 2008. Financial support from the Bloomberg Philanthropies beginning in 20072 and the Bill and Melinda Gates Foundation beginning in 20083 have supported MPOWER-based tobacco control efforts in more than 100 LMICs. As part of the Bloomberg Initiative to Reduce Tobacco Use, the WHO published five reports4–8 documenting the status of the MPOWER measures worldwide.
A previous study9 estimated that over seven million deaths were averted globally as a result of countries adopting ≥1 MPOWER measures between 2007 and 2010. This paper uses newly released data to update the prior analysis to estimate SADs averted as a result of country progress from 2007 to 2014. Although the evidence base establishing the effectiveness of tobacco control policies is vast, no previous studies have systematically estimated the impact of recent progress in tobacco control on population health using SADs averted as a marker of policy impact. Such an estimate is critical, as it allows countries that have not yet put life-saving tobacco control measures in place to better understand the potential benefits for population health.
Assessment of incremental policy change
The WHO reports4–8 provide the status of each nation's tobacco control policy for 2007, 2008, 2010, 2012 and 2014. From these reports, a list of nations that had adopted a highest level policy (ie, placed in a highest level P, O, W or E category or meeting the R requirement that tax is >75% of price, with M providing current smoking prevalence) between 2007 and 2014 was created. To ensure accuracy, we had the list of nations adopting highest level measures confirmed by representatives of the WHO Tobacco Free Initiative. In our previous study,9 we developed a list of nations that met the highest level policy by 2010. In this study, we reviewed 2007 data for any corrections in classifications (as indicated in the online 2015 MPOWER Report) and compared most recently available data (for the year 2014 from the MPOWER 2015 Report) to the corrected 2007 data.
For smoke-free laws, cessation interventions, health warnings and marketing bans, the classifications are based directly on the MPOWER evaluations. For raising taxes, cigarette prices from the MPOWER reports for 2008, 2010, 2012, 2014 and 2015 are adjusted for inflation and translated into effect sizes via an equation dependent on price elasticities.10
Effect size parameters
To examine the impact of highest level MPOWER measures implemented, we applied effect sizes derived from the SimSmoke tobacco control model, which are based on the advice of expert panels and published literature reviews. Since effect sizes are based primarily on policy evaluation studies from HICs, the effects sizes are adjusted by an ‘urban adjustment’ factor (measured in term of per cent urban9 ,11) that captures the ability to reach the population (through smoke-free and cessation policies), and an ‘awareness’ factor (measured in term of whether a low-income, medium-income or high-income nation in 2007)9 ,11 that reflects the potential to affect attitudes at earlier stages in the tobacco epidemic. A parameter based on tobacco control expenditures per capita is also used to calculate publicity effects for smoke-free and cessation interventions. The effect sizes are measured in relative terms as the absolute reduction in smoking prevalence relative to its initial level. We also provided credible ranges for the effect sizes based on a previous review12 and as applied in previous SimSmoke analyses.11 ,13 Policies and effect sizes are shown in table 1 and described in greater detail in our previous paper.9
Effect sizes used account for differing levels of incremental change in policy between 2007 and 2014 (as indicated in the 2015 MPOWER Report), such that countries improving to highest level measures from previously low-level measures show a greater effect size than countries improving from medium-level measures.
The reduction in smokers and smoking-attributable deaths
Smoking prevalence (crude) rates by gender were obtained from WHO reports, measured as smoking (every day and someday) of any tobacco product (including kreteks and bidis). Prevalence estimates were chosen covering a broad age group (preferably ≥age 15) and from surveys closest to 2007 for policies implemented in 2007–2010, for 2010 for policies implemented in 2010–2012 and for 2012 for policies implemented in 2012–2014. Similarly, population estimates (ages 15 and above) are for 2007, 2010 and 2012 for the three groups.14 The smoking rate was multiplied by the population to obtain the number of smokers.
Applying the policy long-term effect size to the number of smokers, we calculated the reduction in the number of smokers. The number of SADs of smokers alive prior to the policy being implemented was calculated by applying an algorithm based on Doll et al 15 and confirmed by later US studies16 ,17 suggesting that half of all cigarette smokers die prematurely from smoking. However, this method does not correct for the higher mortality rate of former relative to never smokers at higher ages. Using US data,18 the ratio of the former smoker mortality rate to the current smoker mortality rate by age and gender was weighted by the number of smokers by age and gender and then summed over ages. This method yielded a former smoker correction factor of an 18% reduction.
Since the number of smokers is measured using the population and the smoking prevalence for the closest available year prior to policy implementation, the number of SADs averted as a result of policy implementation represents the reduction in future SADs of those smokers alive in 2007 for policies implemented in 2007–2010, in 2010 for policies implemented in 2010–2012 and in 2012 for policies implemented in 2012–2014.
For the countries adopting highest level MPOWER measures between 2007 and 2014, table 2 presents year of change, nation income status, initial smoking prevalence by gender, total number of smokers affected, effect sizes and the effects on the number of smokers and SADs averted. Table 3 contains results aggregated by policy from 2007 through 2014. Overall, 88 countries newly adopted a highest level MPOWER policy by 2014, of which 19 are HICs, 51 are MICs and 18 are LICs.
Between 2007 and 2014, 40 nations (6 HICs, 28 MICs, 6 LICs) adopted highest level smoke-free measures. Effect sizes range from −0.4% in Pakistan (due to low enforcement, but worksites and public places laws already in place) to −12.3% in Chile (with few restrictions in 2007 and high enforcement). Of the total 13 198 178 smokers alive in all countries before the policies were implemented, 5 411 253 fewer premature SADs are projected in future years as a result of countries reaching the highest level for smoke-free air, with 2 353 142 fewer SADs due to policies implemented in 2012–2014, 1 749 572 fewer SADs due to 2010–2012 policies and 1 308 539 fewer SADs due to 2007–2010 policies.
Fourteen nations (nine HICs, five MICs) adopted highest level cessation interventions (including quitlines, availability of pharmacotherapies and healthcare provider treatments in all places) from 2007 to 2014. Effect sizes range from −1.8% in Malta to −6.3% in Turkey. A total of 1 495 737 fewer future SADs are projected, with most due to policies between 2007–2010 and 2010–2012.
Thirty-three nations (3 HICs, 24 MICs, 6 LICs) adopted highest level health warning (bold and graphic, and covering at least 50% the package) measures. Effect sizes range from about −2.0% in Canada and Singapore (with strong warnings in 2007) to −8.0% in Sri Lanka, Namibia, Philippines, Turkmenistan, Fiji, Samoa, Solomon Island, Vanuatu and Costa Rica. A total of 4 141 918 fewer future SADs are projected.
Twenty-two nations (3 HICs, 14 MICs, 5 LICs) adopted highest level tobacco advertising bans (on all direct and indirect advertising) with enforcement from 2007 to 2014. Effect sizes range from −1.7% in Guinea (having already had strong laws in 2007) to −13.0% in Turkey. In total, 3 766 340 fewer SADs are projected, of which 2 058 906 deaths are projected due to 2012–2014 policies and 1 504 970 due to 2010–2012 policies.
From 2007 to 2014, 20 nations (6 HICs, 12 MICs, 2 LICs) raised cigarette taxes to 75% of price. The effect size is as high as −40.6% (Latvia), but Cuba, although reaching the 75% goal, increased the tax rate less than the amount needed to compensate for price inflation. In total, 6 976 676 fewer future SADs are projected, of which 4 384 644 deaths are projected from 2012 to 2014 policies.
Table 3 presents the effect on the number of smokers and SADs of countries adopting the highest level POWER measures individually and combined across all 88 countries, and sensitivity analysis of policy effect sizes for SADs. From 2007 to 2010, 31 of the 88 countries implemented at least one highest level measure (22% implemented >1 policy) reducing SADs by 5 336 580. Smoke-free policies and taxes played a prominent role. Between 2010 and 2012, 46 countries implemented highest level measures (17% implemented >1 policy) reducing SADs by 5 365 602 SADs. Smoke-free laws and marketing bans played a particularly large role. Between 2012 and 2014, 32 countries implemented highest level measures (16% implemented >1 policy) reducing SADs by 11 089 741 SADs. Much of the increase in SADs averted from 2012 to 2014 was due to the implementation of highest level measures in four large countries with high smoking rates: Bangladesh (warnings, taxes), Philippines (warnings), Russian Federation (smoke-free air, advertising) and Vietnam (warnings). For all highest level measures implemented between 2007 and 2014, the model projects 53 151 033 (range 30 829 587–75 472 479) fewer smokers, averting an estimated 21 791 924 SAD (range 12 640 131–30 943 716) future deaths among smokers alive before the policies were implemented.
As a result of the highest level MPOWER measures adopted between 2007 and 2014, the number of smokers worldwide is projected to decline by about 53 million, translating to almost 22 million premature SADs averted. This reflects an estimated 16 million additional SADs averted due to policy adoption from our previous analysis spanning 2007–2010.9 The most SADs were averted, due to adoption of increased cigarette taxes, closely followed by comprehensive smoke-free laws, marketing bans, health warnings and cessation treatment programmes. Cessation treatment programmes had the fewest number of countries meeting the highest level policy criteria between 2007 and 2014. Three countries, Brazil, Thailand and Turkey, all MICs, stand out for their efforts in adopting all or nearly all of the highest level MPOWER measures. Of particular note, the number of SADs averted has been increasing over time, with 13 million estimated for 2012–2014. This increase is due primarily to an increased number of countries implementing policies and countries with large populations (eg, the Russian Federation, Viet Nam) implementing policies.
Although 88 countries adopted one or more highest level MPOWER measures between 2007 and 2014, nearly half of the world's population remains uncovered by even a single MPOWER policy. We developed models for the three largest MICs, China, India and Indonesia. We found that, if those three countries alone adopted the complete set of highest level MPOWER measures, the number of SADs of smokers alive in those countries today would be reduced by about 140 million. In particular, few LMICs have met the tax threshold corresponding with the highest level group (>75% of pack price), where this policy has the potential not only to increase cessation among current smokers, but also to reduce initiation of new smokers.19
This model represents a critical step forward in the ability to estimate the public health impact of tobacco control policies. We use data from WHO reports and an extensively validated statistical model11 ,13 ,20–25 to estimate the impact of MPOWER measures—such an assessment provides important, quantitative evidence demonstrating the magnitude of recent tobacco control policy progress globally. In addition, our comparison of the magnitude of the effects from the abridged model used here yielded results consistent with the complete SimSmoke models for nine countries.9
Despite these strengths, the findings should be interpreted in light of the limitations of this work. The estimation method is based on the number of deaths of smokers alive in a particular year and does not incorporate dynamic aspects of the effects of changing demographics and smoking rates and the effects of policies over time. Our estimates exclude smokers who may have initiated after the initial year in the absence of strong measures, thereby potentially underestimating the future effects of policies. However, potentially offsetting these benefits of policy implementation are those smokers who were included and may have quit in future years despite an absence of such policies. In addition, the policy effect sizes have almost exclusively been developed based on policy evaluations directed at cigarettes. Countries with significant use of other tobacco products, such as kreteks, bidis, smokeless tobacco and waterpipe, and weak policies towards the use of these products may experience reduced health benefits through the substitution of these products for cigarettes.
The effects of individual policies are assumed to be independent of each other. This method does not take into account either overlapping or synergistic effects of policies. Knowledge of these effects and how they may depend on the sequence of policy implementation is limited.
The effects of MPOWER policies depend on their specification. The effect sizes for smoke-free and marketing restrictions reflect compliance measures based on enforcement measures, whereas compliance may depend on antitobacco norms reflecting complete implementation of smoke-free and marketing measures as well as other policies. Compliance is not considered for the health warnings and tax (eg, through smuggling) measures. In modelling the effect of tax/price policies, we have taken into account the cost of living. However, as inflation-adjusted incomes rise, some of the effects, especially in low-income and middle-income nations, may be eroded. Finally, the MPOWER measures include only demand-oriented policies, and not supply oriented policies, such as youth access, smuggling and product content regulation.
Our analysis depends on estimates of relative risks of mortality based on data for high-income nations. Nevertheless, relative risks in LMICs, especially in many of the larger countries that are approaching high-income status (eg, Brazil, Mexico, Russian Federation and Turkey), are likely to increase as the intensity and duration of smoking increase and non-smoking-related risks decline.28 If reduced relative risks found in LMICs26 ,27 are applied, 14 551 472 fewer future SADs (range 8 422 560–20 680 385) are still projected as a result of all countries reaching highest level measures. Furthermore, recent studies for high-income nations29–32 find that two-thirds, rather than our estimate of half of smokers, die from smoking in the USA, suggesting that our estimates may be conservative for high-income nations. The estimated SADs also assume that those who quit have the risks of never smokers. However, policies will increasingly reduce smoking rates at younger ages, leaving relative risks close to those of never smokers. In addition, the public health implications go beyond the outcomes examined in this study to include improved quality of life and reduced healthcare costs and productivity loss, harms to the fetus and young children resulting from maternal smoking33 and the additional deaths to non-smokers from reduced exposure to secondhand smoke.34 Thus, the estimate of SADs averted captures only one aspect of the potential positive impact of MPOWER policy adoption on population health. In addition, our analysis only considers outcomes and does not consider cost-effectiveness; some (eg, taxes) may be cost-saving, while others (eg, cessation treatments) have costs, but are still cost-effective.35 ,36
A final limitation is that our estimates do not include the effect of adopted policies that have stronger components but do not qualify as highest level in the WHO reports. We tabulated cases of countries improving, but not meeting the highest level measure for the time period 2012–2014. We found 64 cases of these improvements averting 5.5 million deaths for that time period, suggesting that our estimates provided above that do not include improvements are conservative.
In conclusion, we found a substantial projected impact on SADs resulting from the adoption of highest level MPOWER measures between 2007 and 2014. With an estimated 22 million SADs averted, our findings show the enormous potential to reduce premature mortality by implementing evidence-based MPOWER tobacco control measures in those countries that have not yet implemented highest level measures.
What this paper adds
A previous paper estimated that implementing highest level MPOWER policies between 2007 and 2010 averted about seven million premature deaths.
This paper extends the previous analysis to include 2007 through 2014.
In total, 88 countries adopted at least one highest level MPOWER policy, resulting in almost 22 million fewer premature deaths. The largest number of deaths averted was due to increased cigarette taxes (7.0 million), followed by comprehensive smoke-free laws (5.4 million), large graphic health warnings (4.1 million), comprehensive marketing bans (3.8 million) and comprehensive cessation interventions (1.5 million).
Twitter Follow Darren Mays @darren_mays
Collaborators Jennifer Ellis, Bloomberg Philanthropies.
Contributors DTL conceived of the idea, managed the study, wrote the initial draft and revisions. ZY and YL collected the data, conducted the analysis and helped to revise the paper. DM helped to write and revise the paper.
Funding DTL received funding from Bloomberg Philanthropies through the International Union against Tuberculosis and Lung Disease to conduct this study. The funder helped in the collection of data and interpretation of data. DTL had access to all data in the study and had final responsibility for content of the article and the decision to submit for publication. DTL has also received funding from the Cancer Intervention and Surveillance and Modeling Network (CISNET of DCPS, NCI under grant U01-CA97450-020) for general development of the SimSmoke model and from the National Institute on Drug Abuse, under grant R01DA036497 to disseminate results. Preparation of this publication was also supported in part by a grant from the National Institutes of Health (NIH) and the Food and Drug Administration Center for Tobacco Products to DM (K07CA172217).
Disclaimer The content is solely the responsibility of the authors and does not necessarily represent the official views of the funders.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement We will share the data and programmes used in the article.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.